Lines of code revisited

After a comment by Matt Hamilton on Twitter about the lines of code in my website project I listed in my previous weblog entry, I decided to dive in a bit deeper.

Matt wrote today:

so, same order of magnitude of LoC and resources, but #django provides order of magnitude (or two) less functionality than #plone?

First of all, I think I should start by explaining better what I measured. Both my Plone website and the current Django site are managed via buildout. So when I wanted to count the line numbers of the files under source control, I started by checking out the buildout, and making sure there were no files related to the version control system (.git or .svn directories). Then I simply ran “find . | xargs wc -l” from the buildout directory.

This lead to these numbers:

New numbers

In hindsight this approach was a bit too simplistic. So for this second attempt I took another approach.

In the case of the Plone site I counted the stuff needed for the buildout and my own products to customize the way Plone or installed add-ons behave or look like (basically a policy and theme product).

For the Django that translates in applications that provide the blogging functionality, configuration of the Django project, the templates and CSS and, again, buildout related files.

In both cases I’ve removed unneeded files (like HISTORY.txt, LICENSE.txt, et cetera) and stuff I didn’t use anymore.

The score now:

Wow I didn’t see that coming. So first of all there apparently was was more cruft in the custom code for my Plone site than I figured. And for Django I need more custom code to end up with only the functionality I actually used.

That last conclusion isn’t surprising actually: I decided to build the blogging related code myself on top of Django, instead of reusing a ready-made application.

So my conclusion now is that Matt’s comment was an understatement: I wrote more custom code and ended up with less functionality…